How Much Should You Risk Per Trade (Without Ruin)?

You’ve probably asked it the wrong way round: “How much should you risk per trade to grow faster?”

The real question is, “How much can you lose before you’re finished?”

One of those leads to compounding. The other leads to donating your account to someone with a calculator.

Why “how much to risk per trade” is the wrong starting point

Most people start with a target.

“I want 5% a month, so I’ll risk 2–3% per trade.”

That’s backwards. Returns are an outcome of edge and opportunity; your risk per trade should be anchored to the worst beating your system can hand you without kicking you out of the game.

The market doesn’t care that you fancy 5% this month.

So the correct chain is simple:

• Define the maximum drawdown you can both survive and keep trading through
• Estimate your system’s edge and likely losing streaks
• Back out how much to risk per trade so those streaks don’t hit that drawdown

Notice what’s missing: “what some bloke on YouTube risks”.

Start with ruin: what drawdown can you actually survive?

Forget how brave you feel on a good day.

Think about the point where you would stop trading the system. Either because the capital is too low, or because your head has gone.

For most people, that number is a lot smaller than they admit. Many fold somewhere between a 25–40% drawdown, even if they swear they could sit through 60% “because it’s long term”.

So pick a hard stop.

Example bands (hypothetical, not prescriptions):

• Cautious: “If I hit 20% drawdown on this system, I’m out.”
• Moderate: “30% is the pain line.”
• Aggressive: “I can live with 40–50% on this sleeve of capital.”

That’s your survival constraint. Risk per trade lives inside it.

This is the risk-of-ruin mindset — not literally to zero, but to the point you realistically abandon the system.

Edge first, bravado second

Next problem: you cannot choose how much to risk per trade without some idea of edge.

That means at least an approximate win rate and average win/loss size. Nothing fancy, but it needs to be honest, not curve-fitted fantasy.

If you haven’t thought about expectancy properly, read the trade-off in /blog/win-rate-vs-risk-reward-the-trade-off-nobody-explains-2, then come back.

Quick hypothetical:

• Win rate: 50%
• Average winner: +2R
• Average loser: -1R

Expectancy per trade = 0.5×2 − 0.5×1 = +0.5R.

So you make half a risk unit, on average, per trade.

If R is 1% of your account, average gain is +0.5% per trade before costs and slippage. If R is 4%, average is +2% per trade. Tempting.

Until you look at the drawdowns and strings of losses that 4% creates.

Why the win rate won’t save you

People love win rate.

“My system wins 70% of the time, I can risk 3–5% a trade.”

Except that even with a high win rate, losing streaks happen. You don’t get to negotiate with the binomial distribution.

Rough rule of thumb from probability theory:

The longest losing streak you’re likely to see over N trades is on the order of log(N) divided by log(1/(1−p)), where p is the win rate.

You don’t need to memorise that. You just need to internalise that a 60–70% win rate still throws 6–10 losers in a row over a few hundred trades, quite happily.

Take a simple example for intuition:

• Win rate: 60%
• Across 500 trades
• Simulated a few thousand times (Monte Carlo style)

You’ll commonly see losing streaks of 7–9 trades somewhere in that run, sometimes worse. We go into why your backtest underestimates this in /blog/what-is-monte-carlo-simulation-in-trading.

Now layer risk per trade on top.

When “just 2% per trade” quietly kills you

Answering “how much should you risk per trade” with “2%” is popular.

It sounds professional. It is, in some cases, still aggressive.

Suppose you hit a 10-trade losing streak. At 2% per trade, the loss is not 20%.

It’s worse, because 2% is of a shrinking base.

Drawdown after n losses at fixed-fractional risk r is: Equity = (1−r)^n.

At 2% risk (r = 0.02) and 10 straight losses: Equity = 0.98^10 ≈ 0.817. That’s an 18.3% drawdown.

At 4% risk?

Equity = 0.96^10 ≈ 0.665. Roughly a 33.5% drawdown from a single nasty streak.

At 0.5% risk, the same 10 losers gives you 0.995^10 ≈ 0.951. About a 4.9% drawdown. Same system, same bad run, totally different survival odds.

The system didn’t change. Only your ego did.

Fixed-fractional risk: boring, and that’s the point

Once you know your maximum acceptable drawdown and a realistic worst-case losing streak, you can work back to a sensible fixed-fractional risk per trade.

This is the unglamorous method where you risk a fixed percentage of current equity each trade. Position size floats with account size, but the percentage stays the same.

We compare it against fixed-lot sizing in /blog/fixed-fractional-vs-fixed-lot-which-sizing-really-compounds, but the short version: fixed-fractional is what actually compounds and self-defends in drawdowns.

Example approach, purely hypothetical:

• You decide you never want a typical losing streak to cost more than half your acceptable drawdown
• Max acceptable drawdown: 30%
• You want a “normal bad streak” to be ~15% or less
• You judge a realistic worst streak as 10 losers (again, via testing and Monte Carlo, not optimism)

Plug it into (1−r)^n ≥ 0.85, where n = 10 and 0.85 is 1 minus your 15% tolerance.

So (1−r)^10 ≥ 0.85.

Take the 10th root: 1−r ≥ 0.85^(1/10) ≈ 0.9846.

So r ≤ 1.54% per trade.

If you want to be more conservative and keep that typical streak to, say, 10%, you get r closer to 1%.

This is how you should be answering “how much to risk per trade”.

With a calculator and a drawdown constraint, not with your favourite integer.

Risk per trade and risk of ruin

The nasty cousin of drawdown is risk of ruin.

Not glamorous, but essential.

Risk of ruin is the probability you hit your “I’m done” drawdown at some point in the future given your edge and risk per trade.

There are various formulae; the exact maths is messy and sensitive to assumptions.

The concept, though, is clean: for a given system expectancy and variance, raising risk per trade ramps up the chance you’ll blow through your pain threshold before the long-run edge saves you.

Even a positive-expectancy system can have a terrifying risk-of-ruin if you size like a hero.

A rough practical approach:

• Take your trade history or backtest
• Parameterise risk per trade (e.g. 0.25%, 0.5%, 1%, 2%)
• Run Monte Carlo resamples to generate many randomised paths
• Measure how often equity drops to your “ruin” level at each risk

You’ll usually discover that somewhere between 0.25–1% risk per trade, risk-of-ruin at sane horizons collapses to something you can live with.

That’s the boring truth nobody puts on Instagram.

Account size, instruments and minimum lot constraints

Everything so far assumes you can actually trade the precise risk per trade you choose.

With small accounts and instruments like gold (XAUUSD), that often isn’t true.

Example: if the smallest trade size is 0.01 lots and the instrument is volatile, the monetary swing per trade can be large relative to a tiny account.

So on a £500 account, a 0.01 lot gold trade might easily be equivalent to risking more than 1–2% per trade, depending on your stop distance.

You can’t “risk 0.25% per trade” if the smallest allowable position size is already 3% of your equity at your stop.

That’s not a position sizing philosophy problem, it’s a capital adequacy problem.

So part of working out how much to risk per trade is checking whether your desired percentage risk is even implementable with your balance, your broker’s contract sizes and your stop placement.

If not, your options are brutally simple:

• Deposit more so that 0.01 lots is a smaller percentage of equity
• Trade less volatile instruments or higher timeframes with wider, more meaningful stops
• Accept that you’re running high percentage risk and treat it as speculative, not “position sized”

Reality doesn’t bend just because the spreadsheet says 0.5% is optimal.

Why “all-in on the best setup” is still nonsense

A popular discretionary idea: risk more on your “A+ setups”, less on the rest.

Nice story; zero evidence usually.

Unless you’ve got statistically significant data that your A+ setups genuinely have higher expectancy, what you’re actually doing is concentrating risk into a tiny sample of trades that feel good.

The maths is unforgiving.

If your labelling of “A+” is noisy (it will be), you’re just increasing variance: when your intuition misfires, losses are magnified.

This is why most systematic traders keep risk per trade flat in percentage terms and let the edge express over many trades, not three heroic ones.

If you want to ladder risk, do it systematically and test it: rules, thresholds, samples, the lot.

Not based on how much you like the candle pattern.

Automation doesn’t remove the risk. It removes the tampering.

There’s a separate psychological mess behind how much to risk per trade.

You can choose a beautifully conservative 0.5% per trade, then quietly override it after three winners because you’re “trading with house money”.

Or slash it to 0.1% after a normal drawdown because “the system’s broken”. We talk about that cycle here: /blog/why-traders-abandon-winning-systems-at-the-worst-time.

Automation can’t make a bad risk plan good.

But it can enforce the plan you actually signed up to, not the one your amygdala invents on a bad Tuesday.

Once you’ve done the work on edge, losing streaks and drawdown tolerance, automating position sizing and execution is mostly about getting yourself out of the way.

No martingale. No grid. No “double after loss because surely it turns soon”.

Just a fixed-fractional risk per trade, chugging through the distribution.

Utterly boring. That’s the idea.

Putting numbers on it: a practical range

So what does all this mean in practice for how much to risk per trade?

Assuming you have:

• A system with positive expectancy (tested, not imagined)
• Hundreds of trades of data or a solid backtest
• A realistic maximum drawdown tolerance in mind

You then run scenarios, either analytically or via Monte Carlo, and look at these for different risk-per-trade levels: maximum and median drawdown, length of drawdowns, risk of hitting your pain line.

Across many such hypothetical exercises, you’ll keep circling back to roughly the same bands:

• 0.25–0.5% per trade: conservative, low risk of ruin, psychologically easier, slower equity curve
• 0.5–1% per trade: moderate, a common sweet spot for diversified, robust systems
• 1–2% per trade: aggressive, drawdowns and streaks start to bite; suitable only if both edge and stomach are strong

Above ~2% per trade, for most genuine systems (not fantasy ones), risk-of-ruin and drawdown depth accelerate faster than the extra return is worth.

Kelly criterion gets mentioned a lot in this context.

Yes, there is a formula that maximises long-run growth for a known edge and variance. No, you should not blindly bet full Kelly in real trading where both are uncertain and changing.

Most serious players treat “fractional Kelly” as an upper bound and stay at a fraction of it. That fraction often happens to land in the 0.25–1% world once translated to percentage-of-equity risk per trade.

Different systems, different sleeves

One more nuance: you don’t have to pick one global number for how much to risk per trade for everything you do.

If you run several uncorrelated systems — say trend-following on FX, a mean-reversion index strategy, and something on gold — you can size each system separately within an overall portfolio risk limit.

The questions then become:

• What’s the worst combined drawdown I can tolerate across all systems?
• How correlated are they in stress periods (not just in a backtest average)?
• Given that, how much risk per trade can each system run so that the portfolio stays inside that drawdown band?

This is portfolio risk management, not just single-strategy bravado.

And yes, systems on more volatile instruments or with lower win rates will often get a lower percentage risk per trade allocated to them.

Not because they’re “worse”. Because their distribution of outcomes is lumpier.

The simple rule that survives contact with reality

If you strip out all the maths, what’s left is this:

• Decide the deepest drawdown you can survive, both financially and psychologically
• Estimate your system’s losing streaks from real data, not hope
• Choose a fixed-fractional risk per trade so that a realistic worst streak lives comfortably inside that drawdown
• Automate it so you don’t move the goalposts mid-game

That’s how you answer “how much should you risk per trade”.

Not by what you want to make this month, but by what you can afford to lose and still be here next year.

Trading involves real risk of loss. You can lose some or all of the capital you put at risk.

Where to go next

If you’ve never quantified your losing streaks or modelled drawdown properly, start there.

/blog/what-is-monte-carlo-simulation-in-trading will show you why your backtest is lying about how smooth the ride will be.

/blog/loss-aversion-in-trading-why-you-cut-winners-and-hold-losers covers why your brain hates drawdowns even when the maths is fine.

Then sanity-check your current risk per trade against those numbers.

If you’re surprised by how aggressive you’ve been, good. Better surprised on a spreadsheet than in a margin call.

Related reading

• What Is Sharpe Ratio Trading (And Where It Misleads)?
• What Is System Decay In Trading (And When To Retire One)?
• Loss Aversion in Trading: Why You Cut Winners and Hold Losers

Start your free 14-day ArcisTrade demo →

Watch real automated strategies run in a demo, with fixed-fractional risk and no martingale or grid under the hood.

See the winners and the losers, the swings and the drawdowns, before you put any capital at stake.

P.S. Use the demo to stress-test your own risk per trade ideas against how actual systems behave, not how spreadsheets behave.